Principles
OTD is built on four principles. Everything else in the framework follows from them.
1. Empty your head.
Get everything out of your mind and into the system. As long as you're holding tasks in your head, no system will work.
Why this can't be compromised
Your brain is not storage. It's a processor. When you use it as storage, you get two problems at once: the stored items degrade (you forget things) and the processor degrades (you can't focus because you're also trying to remember).
A trusted external system fixes both. You stop forgetting. You stop worrying about forgetting. The mental overhead of remembering that you need to remember something (that specific tax on your attention) disappears.
The insight is simple but easy to underestimate: the volume of things competing for your attention only grows over time. A system you actually trust is what makes focus possible.
What "the system" means
The system doesn't have to be sophisticated. It has to be:
- Trusted: you believe it holds everything, so you stop mentally tracking backup copies
- Consistent: you actually use it, every time, for every input
- Reviewable: you look at it regularly enough to keep it current
If any of these three break down, your brain starts re-absorbing tasks. That's the failure mode.
2. Curate your day, design your life.
The system fills Today. You curate it.
Active Flows surface their front Actions automatically. Your role each morning is to look at what surfaced and confirm it's the day you want: accept, swap, drop, occasionally add. Curation can take thirty seconds or ten minutes. Both are valid.
When even thirty seconds feels heavy, run the auto-filled slate as-is and execute. The agency you don't spend daily, you spend at Weekly and Quarterly Review. The principle is conservation of agency, not prescription of cadence.
Four failure modes a productivity system has to guard against
These pull in opposite directions, which is why no single rule covers them all:
- Queue-executor failure: The tool decides everything. You execute its queue. Busy, but going nowhere. Long-term cost: loss of agency.
- Decision-fatigue failure: Every Action requires deliberate selection from a blank slate. You burn willpower before doing the work. Mornings collapse into staring. Long-term cost: avoidance and abandonment of the system.
- Drift failure: Neither the system nor you are actively choosing. Days pass mechanically. A year goes by without movement on what mattered.
- Approval-theater failure: The decision points exist, but they stop deciding. Work returns faster than you can judge it, so approving becomes a reflex. You are formally in control and practically not. Long-term cost: the illusion of agency, which is worse than its absence because it hides the drift. Diagnostic: your approval rate sits near 100%.
A rule like "build Today from scratch every morning" only addresses the first failure. For people prone to the second (anyone with too many wants, too many active commitments, or mornings already heavy with executive load), that rule actively makes things worse. They abandon the system within weeks, not because they're undisciplined, but because the ritual is mismatched to the failure mode they actually have.
OTD's answer is automatic surfacing plus deliberate curation. The system carries the cognitive load of "what's next?" Within each Flow, sequencing is automatic. Across Flows, the front Action of each surfaces into Today without effort. What remains for you is confirmation. And only when it's worth your attention.
Which failure mode are you prone to?
Most people lean toward one. Knowing which one shapes how you should run OTD.
You're prone to Queue-executor failure if:
- You've abandoned other productivity apps because they "told you what to do" and it felt empty
- You finish busy days unsure whether anything you did actually mattered
- You instinctively distrust auto-prioritization features
- You think of yourself as deliberate, methodical, or strategic
The morning ritual is your friend, not overhead. Recommended cadence: daily curation kept honest, with lighter Weekly Review. Resist letting auto-fill carry you for too long without questioning it.
You're prone to Decision-fatigue failure if:
- You have many things you want to do, too many to honestly choose between every morning
- Your mornings already feel heavy before work starts (executive function spent on family, health, other domains)
- You've abandoned other productivity apps because the daily setup felt like another job on top of the work
- You're often in the state of "I have so much to do I can't start anything"
This is the profile of the over-wanter. The daily ritual that helps the first profile actively harms you. Recommended cadence: light daily curation (often just executing the auto-filled slate is right), with heavier Weekly and Quarterly Review. Spend agency at the cadence where you have energy, not the one productivity culture prescribes.
You're prone to Drift failure if:
- You're in a routine-heavy phase of life and most days feel similar
- You can't easily remember the last time you decided something about your active set
- You can't say in one sentence what you're working toward this quarter
- You've been executing a lot recently but can't articulate the direction
Recommended cadence: whatever daily mode you prefer. But Weekly and Quarterly Review are non-negotiable, and they need to actually decide things. Showing up to Review without making decisions is just performing the ritual.
You're prone to Approval-theater failure if:
- You delegate to AI often, and results come back faster than you read them
- You cannot remember the last time you rejected something that came back
- You approve in batches to clear the queue rather than to make a judgment
- You'd have to check the log to know what changed in your system yesterday
This profile is new. It barely existed before delegation became cheap. The other three are about choosing what to do; this one is about accepting what came back. Recommended cadence: lower your Review budget until judging feels real again, narrow how many agents can write, and put friction on returns that change or remove existing work. Approving less per day is the fix, not approving faster.
You can be prone to more than one. The most common combination is Decision-fatigue plus Drift: the over-wanter profile that defaults to executing whatever surfaces because choosing has become exhausting, and so the active set silently drifts. The answer for this profile is the same: anchor on Weekly Review, take it seriously, and let daily run on auto. The daily ceremony was never going to fix this. Review is.
Where agency lives
Whatever your morning looks like, OTD requires you to remain the author at some cadence:
- Quarterly: what themes and Goals matter this season
- Weekly: which Flows are active, what's stuck, what's closing
- Activation: Principle 3 trade (what's closing to make room for what's opening)
If your daily curation is light or skipped, Weekly and Quarterly Reviews become non-negotiable. That's where the steering happens. Skip them too and you slide into Drift.
If your daily curation is heavy, Reviews can be lighter. Daily engagement carries some of that weight.
The freedom is symmetric: agency you don't spend daily, you spend periodically. That symmetry is what makes the system livable for both over-deliberators and over-wanters.
The difference between sequencing and designing
OTD provides automatic sequencing within each Flow: the first uncompleted Action surfaces in Today, so you never have to think "what's next?" inside a single stream of work.
Today's auto-fill extends this across Flows: the front Actions of all your active Flows arrive together each morning, ready to execute.
Designing your life happens at a higher level: which Goals you're pursuing, which Flows you've activated, what's closing to make room for what's opening. Those are human judgment calls. They depend on your energy, your context, your sense of what matters. A system can surface signals. It can't make those choices for you.
What this means for AI
AI in OTD is an executor. It can populate Today, suggest swaps, surface Stuck Signals, and carry Delegate / Await / Review cycles. What it cannot do is choose your themes, your Goals, or what closes when you activate something new.
What comes back from AI enters through Review, never directly (Principle 4). Filling your day is delegable. Accepting the result is not.
The boundary between machine fills your day and machine decides your life is the boundary this principle protects.
3. Close before you open.
To activate a new Flow, decide what's closing. Complete it, move it to Someday, demote it to Wishlist, or delete it.
The asymmetry this corrects
Principles 1 and 2 handle input and execution. They don't handle a quieter failure: the active set growing without limit.
Starting a new commitment is structurally cheap. It enters Inbox, gets processed, becomes a Flow. The friction is low and the system encourages it.
Closing a commitment is structurally expensive. You have to look at a Flow and admit one of four things: I finished it, I'm parking it, I'm not really doing this, or it's gone. Each of those takes a deliberate decision.
When the easy direction (start) is asymmetrically easier than the hard direction (close), the active set expands without bound. You accumulate Flows you're "going to get to." Today becomes too crowded to finish. The completion signal breaks. The system you trusted starts feeling like a backlog you're managing.
Principle 3 forces symmetry: every new activation pairs with a closure decision.
What "close" means
Four ways to close, in order of weight:
- Complete: the Flow finishes naturally. Best case.
- Move to Someday: you're not doing this now, but you'll reconsider later. The intention to act remains.
- Demote to Wishlist: you're not committing. It's a possibility you're keeping, not an obligation.
- Delete: the work is no longer relevant. Acknowledge and remove.
Closing isn't failure. It's the maintenance work that keeps the active set inhabitable.
How this works with parallel Flows
OTD is designed for parallel work. "Sequential inside, parallel across" is not negotiable. Multiple Flows advancing simultaneously is the core of the framework. Principle 3 does not restrict parallel execution.
What it restricts is the size of the active set. Five to ten Flows running in parallel is normal. The principle activates at the margin, when you reach for one more. Before adding it, decide what's leaving.
The principle doesn't say "finish one thing before starting another." It says: the active set has a ceiling, and growing past it requires a deliberate trade.
When the principle applies
- New Flow activated from Inbox → identify which existing commitment is leaving. Trade required.
- Moving from Someday or Wishlist into active → this is a new activation. Trade required.
- Stuck Signal flagged a Flow you decided to keep active "with reason" → the act of keeping is its own decision; you're not adding new commitment, you're explicitly renewing one. No trade required.
- A recurring Flow resets → not a new activation. No trade required.
The principle protects the active set from passive accumulation. It doesn't add bureaucracy to natural progression.
Why this isn't just a rule
It would be possible to enforce this as a tool feature: "you can't create a new Flow without archiving one." OTD doesn't prescribe that.
It's a principle because the trade has to be conscious. A tool gate would let you archive any Flow at random just to clear the path. That defeats the point. The decision of what to close is part of designing your active life, just like Principle 2's decision of what you're steering toward. Both are yours to make.
What the principle does is name the moment. When you create a new Flow, the question "what's closing?" becomes a part of activation, not an afterthought.
4. Nothing lands unreviewed.
Work that comes back from Await enters through Review. It is never applied on arrival.
Why delegation needs a return gate
Principles 1 through 3 cover input, agency, and the ceiling of the active set. They share an assumption: what you delegated comes back the way you remember sending it, at a pace you can absorb.
With people that assumption mostly holds. A colleague replies; you read the reply. Delegation is slow and low-volume, so the return takes care of itself.
With AI it does not hold. One agent can return a dozen results in minutes, and most tools apply those results the moment they arrive. The work lands in your system without passing through you.
That breaks Principle 1 from the far end. You still hold everything. But you no longer know which parts you chose. A system you can't distinguish your own decisions in is not a trusted system.
The gate
- Await → Review → in. There is no other path for returned work.
- Review is a decision: accept, reject, or send back.
- Reading needs no gate. Anything that changes your system does.
- Rejections leave a trace. Nothing disappears silently, and nothing arrives silently.
Actor matters
Await and Review carry an actor: human, agent, or rule (see Mode). All three return through the same gate, but they fail differently:
- Human: slow, few, easy to forget. The risk is losing track.
- Agent: fast, many, plausible. The risk is volume outrunning judgment.
- Rule: invisible while it works. The risk is forgetting you wrote it. A rule is only legitimate while you can read it back as your own sentence.
Review has a budget
A gate with no budget becomes a second Inbox. It's the exact failure Principle 1 exists to prevent, arriving through the back door.
Cap how much you review in a day. When the cap is reached, the machine waits, not you: agents stop being able to hand work back until you have room. Unreviewed items age out and surface as Stuck Signals rather than piling up.
What the system cannot promise
Await records that you handed something off. It does not verify that the work was done. A returned result is a claim, and Review is where you weigh the claim. A framework that pretends otherwise turns delegation into faith.
Why this is a principle and not a feature
A tool can implement the gate, and a good tool should. But the gate is worthless if approving is cheaper than thinking (see Approval-theater failure in Principle 2). What makes this a principle is the standard it sets: the final decision is human, at the moment work enters your system. A tool can hold the door. Only you can decide what walks through.
How the four principles interact
Principle 1 creates the condition: a complete, trusted external system that holds all your work.
Principle 2 creates the agency: you remain the author of your active life, daily by light curation, weekly by review, quarterly by direction. The cadence is yours; what's not optional is that somewhere you're steering.
Principle 3 creates the boundary: the active set you're choosing from stays inhabitable over time.
Principle 4 creates the seal: nothing enters the system without your decision, no matter who or what produced it.
Together, they produce a specific outcome: you're never surprised by what you're working on, you never feel like you're just reacting, and the system doesn't quietly bloat into a backlog you can't finish. Principle 1 puts everything in. Principle 2 keeps you in the driver's seat without forcing daily ceremony. Principle 3 keeps the road inhabitable. Principle 4 keeps the road yours. As delegation gets cheaper, it is the principle that stops the system from filling with work you never chose.
That's the core of OTD.
On productivity and guilt
Many productivity systems create a specific kind of harm: the sense that you're always behind, always failing to do enough, always carrying a backlog of things you "should have" done.
OTD is designed against this.
- There is no overdue list. Actions that aren't completed today return to their Flows and surface again tomorrow. They don't accumulate in a pile of guilt. The list doesn't grow. Tomorrow starts fresh.
- "Today is empty" is a real completion signal. Not "it's 6pm," not "I'm tired," not "I ran out of willpower." Today is empty. The Actions you chose this morning are done. That's an objective measure of a completed day, not a comparison against some ideal of maximum output.
- You set the standard each morning. During Daily Review, you choose which Actions go into Today. You're not measured against everything that could possibly be done. You're measured against what you decided to do today. If Today is empty, you succeeded by the standard you set for yourself.
This doesn't mean working less. It means working without the background noise of "I should be doing more." That noise is expensive. It degrades focus, produces anxiety, and eventually leads to abandoning the system entirely to escape the judgment it represents.
A system that ends your day cleanly, with a real signal that today's work is done, is not a luxury. It's what makes the system sustainable over months and years, not just weeks.